Commztnications Peroxy Radical Cyclization as a Model f o r Prostaglandin Biosynthesis
COOH 11
soybean lipoxidase
CH,N,
Summary: Unsaturated lipid hydroperoxides are converted into prostaglandin analogs by free-radical initiators. Sir: The biosynthesis of prostaglandins has been the focus of extensive investigation over the past several years.l A peroxy radical cyclization mechanism leading to intermediate endoperoxides has been proposed for this biosynthesis (Scheme I). The intermediacy of endoperoxides has, in
-€I,
C
C
5
H
l
l
2
3
were assigned by conversion to the ketostearates followed by mass spectral analysis.4b Incubation of 2 (10.8 mg, .0333 mmol) with di-tert-butyl peroxyoxalate (DBPO; 3.22 mg, 0.014 mmol) in 02-saturated benzene at 25' for 16 hr led to complete consumption of 2. The products of the reaction, all of which were more polar than starting hydroperoxide, were reduced with NaBH4 (25 mg, 0.661 mmol in 10 ml of CN3OH) and then converted into the trimethylsilyl ethers for analysis by gas chromatography (GLC). Chromatography of the reduced and silylated product mixture on a 1%OV-17 column at 220' revealed a t least four volatile products with retention times expected for PGF-type products from a C-18 fatty acid.g Combined GLC-MS analysis of the product mixture indicates that two of these four compounds (retention times 9.5 and 10.5 min) are structurally analogous to authentic PGF1,. Prominent fragments of one of these compounds and PGF1, analyzed under identical GLC-MS conditions are presented in Table I.
Scheme I
COOH
\ OOH
HOO
o2
enzyme
fact, been concretely established by the isolation of two such compounds, PGG and PGH, from tissue homogenates which were actively engaged in the production of prostaglandim2 Recently, we reported a method for generating specific peroxy radicals from the corresponding hydroperoxide.3 This method is based on the fact that hydroperoxy hydrogens are readily abstracted by free radicals. Thus, hydroperoxides act as a source of the corresponding peroxy radical when treated with appropriate free radical initiators. We report here the application of this method to a hydroperoxide which should lead to a peroxy radical analogous to
Table I Mass Spectral D a t a (I.P. = 70 eV) of Trimethysilyl Ethers and Methyl Esters of PFG1, and Synthetic C-18 P G F (Retention Time 9.5 min) PGF 10
Polyunsaturated fatty acids are converted into hydroperoxides by a class of enzymes known as lipoxygenases (fatty acid: oxygen oxidoreductase). These enzymes have been identified in extracts from a variety of plant sources including soybeans: potato tubers: and wheat flour.6 With the soybean e n ~ y m e y-linolenic ,~ acid (all-cis-6,9,12-octadecatrienoic acid) could be converted into a 1:l mixture of the w-10 and a-6 hydroperoxides 2 and 3. The reaction was carried out on the ammonium salt of the fatty acid a t pH 7 for 2 min.8 The crude hydroperoxide fatty acids were then converted into the hydroperoxide methyl esters 2 and 3 by reaction with diazomethane. 2 and 3 could be separated by high performance liquid chromatography (HPLC) with 6 f t x in. of Corasil I1 and 0.24% isopropyl alcohol in hexane solvent. Thus, pure 2 could be isolated in an overall 15-20% yield based on starting fatty acid. The structures of 2 and 3
FragmenP
m/e
A
M
496 355 353 335 309 297 295
2 24 100 72
-
90
M - 233 M-251 M - 277 M-289
1.
C-18 PGF
80
62 50
Fragmenta
mte
94
M-90
468 3 55 325 307 281 269 295
2 48 72 61 75
MMMM-
233 251 277 289
100 22
Fragments between 200 and 500 m / e are shown and referred to the largest peak in this region as base. a
The mass spectra of the synthetic PGF's (which have carboxyl side chains smaller than PGFI, by CzH4, 28 m/e) are strikingly similar to PGFI,.~" Every fragment from the synthetic PGF's between 200 and 500 m/e with a relative intensity greater than 20% of base corresponds either t o a fragment of the same mass (loss of the carboxyl side chain) or to a fragment of m/e - 28 (retention of carboxyl side chain) from PGF1,.ll A mechanism consistent with our observations is presented in Scheme 11. tert- butoxy radicals generated by de-
3614
J. Org. Chem., Vol. 40, No. 24, 1975 3615
Communications composition of DBPO abstract the labile hydroperoxide hydrogen3 from 2 giving the peroxy radical. Cyclization of this radical to the endoperoxide followed by reductiod2 leads to PGF-type products. Scheme I1 COOCHB 2
+ t-BuO.
cyclization
C5H,,
-+
.o-0
/
bOH
+o,. +H.
OH
I
OH
4
The work reported here lends support to the notion that prostaglandin biosynthesis is a controlled free-radical reaction.' Other workers have noted the formation of prostaglandins in autoxidizing lipid.11J3 The method reported here has the advantage of producing specific peroxy radicals for study 8.8 compared to the rather random autoxidation format. Acknowledgmunt. This work was supported by grants from NIH and the Army Research Office (Durham). GLC-MS work was carried out a t the Research Triangle Institute. References a n d Notes (1) (a) 8. Samuelsson, Fed. Proc., fed. Am. SOC. Exper. B/ol., 31, 1442 (1972); (b) 8. Samuelsson, H. Granstrom, and M. Hamberg, In Nobel Symp. n2, 31-45 (1987): (c) D. H. Nugteren, R. K. Beerthuis, and D. A. van Dorp, lbld., 45-50 (1987). (2) (a) D. H. Nugteren and E. Hazelhof, Blochlm. Blophys. Acta, 328, 448 (1973); (b) M. Hamberg and B. Samuelsson, Proc. NaL Aced. Scl., USA, 70, 899 (1973); (c) M. Hamberg, J. Svenson, T. Wakabayashl, and B. Samuelsson, lbld., 71, 345 (1974). (3) (a) M. Funk, R. Isaac, and N. A. Porter, J. Am. Chem. Soc., 97, 1281 (1975); (b) see also J. A. Howard, Adv. Free-Radlcal Chem. 4, 49 (1972); (c) see also R. Hlatt, T. MIII, K. C. Irwln, and J. K. Castleman, J. Orp. Chem., 38, 1428 (1968). (4) (a) M. Hamberg and B. Samuelsson, Blochem. Blophys. Res. Commun., 21, 531 (1985); (b) M. Hamberg and 8 . Samuelsson, J. 8/01,Chem., 242, 5329 (1987); (c) M. Roza and A. Franke, Blochlm. Blophys. Acta, 318, 78 (1973); (d) J. P. Chrlstopher, E. K. Plstorlus, F. E. Regnler, and B. Axelrod, lbld., 289, 82 (1972), (5) T. Galllard and D. R.Phllllps, Blochem. J., 124, 431 (1971). (8) A. Graveland, Llplds, 8, 808 (1973). (7) Slgma Blochemlcal Co.; 45,500 unlts mg-l at pH 7.0. (8) N. A. Porter and M. 0,Funk, to be publlshed. (9) Several other more volatlle products were also observed. (10) See, for example, (a) C. J. Thompson, M. Los, and E. W. Horton, Llfe Sclences, 9, 983-988 (1970); (b) S. Nicosla and 01. Qalll, Anal. 810chem., 81, 192 (1974); (c) M. Hamberg, Eur. J. Blochem., 8, 135 (1968); (d) P. Wlodawer and B. Samuelsson, J. Blol. Chem., 248, 5873 (1973). (11) (a) Our mass spectra were compared to those of compounds obtalned by autoxidatlon of methyl a-llnolenate (W. A. Pryor and J. P. Stanley), Fragmentatlon patterns were generally slmllar wlth some exceptlons that may be ewpllcable by dlfferences In the proposed structures. We thank Professcr Pryor for this information before publlcation. (b) W. A. Pryor and J. P. Stanley, Abstracts of Papers, American Chemical Soclety Meeting, Chicago, Ill.,Aug 25-29, 1975, paper ORGN-50. (12) An alternate route to PGF compounds is via lsomerlzatlon of 4 to PGE's, followed by reductlon of PGE to PGF. (13) D. H. Nugteren, H. Vonkeman, and D. A. van Dorp, Recuell, 86, 1237 (1987).
Department of Chemistry Duke University .Durham. North, Carolina 27706 Received August 15, 1975
Ned A. P o r t e r * Max 0.F u n k
A Suggested Mechanism f o r t h e Production of Malonaldehyde d u r i n g the Autoxidation of Polyunsaturated F a t t y Acids. Nonenzymatic Production of Prostaglandin Endoperoxides d u r i n g Autoxidation
Summary: Autoxidation of methyl linolenate is shown to yield materials which give positive tests for both prostaglandin E and malonaldehyde, and it is suggested that both tests respond to prostaglandin-like endoperoxides which can be formed by autoxidation. Sir: When polyunsaturated fatty acids (PUFA) or esters containing three or more double bonds undergo autoxidation, a material is produced which develops color in a sensitive test with thiobarbituric acid (TBA).1-9 This TBA test is the most frequently used index of lipid peroxidation both in vitro and in vivo. Although the TBA-reactive material is frequently referred to as m a l 0 n a l d e h y d e , 5 ~it~has ~ ~ ~been ~~ known for some time3p4 that this material is predominantly nonvolatile; therefore, it is not malonaldehyde, but rather is a nonvolatile precursor of malonaldehyde. In 1962, Holman et aL4 suggested that a five-membered monocyclic peroxide is the nonvolatile malonaldehyde precursor. However, their mechanism does not appear to accommodate all the known facts of TBA-color production during PUFA autoxidation.1 A more attractive mechanism, in our view, irJ one in which the nonvolatile malonaldehyde precursor is a bicyclic endoperoxide analogous to that which is formed in the biosynthesis of prostaglandins.11-2a Figure 1 shows this mechanism as applied to methyl linolenate (18:3).Abstraction of an "internal" allylic hydrogen followed by reaction with 02 leads to peroxy radicals 4 and 5. Radical 4 has a structure which allows cyclization to endoperoxide radicals 9, which are allylic, probably via the oxy-bridged radicals 6. Radicals 9, then, can become converted into endoperoxides 10 or 11.21-23Radicals 4 and 5 also can lead to the conjugated hydroperoxides 7 and 8 which are known products of autoxidations. Our first indication that the nonvolatile malonaldehyde precursor is an endoperoxide came from comparisons of the responses of autoxidized solutions of 18:3to the TBA test and a test developed for prostaglandin E (PGE).1Q*z4~25 The PGE test, which involves rapid formation of absorption a t 278 nm upon addition of alcoholic b a ~ e , ~ ~probably * ~ ~ l ~isd relatively unspecific, but it is believed to convert PGE compounds into conjugated dienones such as PGB.24Since base is known to rapidly decompose secondary dialkyl peroxides to form ketones and alcohols,2e-2s we expected that endoperoxides, if produced in our autoxidations, would be converted by baae into PGE-type compounds. The PGEtype compounds would then react further with base to give PGB-type chromophores and a positive PGE test, It appeared reasonable a priori that endoperoxides could be formed nonenzymatically by autoxidation in our system since the suggested mechanism for their biosynthesis involves a radical cyc1ization;l la0 furthermore, Nugteren et al.2Dahave shown that autoxidation of 8,11,14-eicosatrienoic acid gives prostaglandins. Thus, we hypothesized that endoperoxides are produced on autoxidation of 18:3 and are the precursors of malonaldehyde under TBA-test conditions and PGB under conditions for the PGE test. (Note that 10 and 11 should give malonaldehyde but only 11 should give a PGE test.) Indeed, autoxidized 18:3 does give a PGE test. This is true regardless of whether the oxidation is spontaneous (Le., effected by pure air), or is initiated',' by ozone or NOr. On the other hand, 183, autoxidized under the same condi-
COOH 11
soybean lipoxidase
CH,N,
Summary: Unsaturated lipid hydroperoxides are converted into prostaglandin analogs by free-radical initiators. Sir: The biosynthesis of prostaglandins has been the focus of extensive investigation over the past several years.l A peroxy radical cyclization mechanism leading to intermediate endoperoxides has been proposed for this biosynthesis (Scheme I). The intermediacy of endoperoxides has, in
-€I,
C
C
5
H
l
l
2
3
were assigned by conversion to the ketostearates followed by mass spectral analysis.4b Incubation of 2 (10.8 mg, .0333 mmol) with di-tert-butyl peroxyoxalate (DBPO; 3.22 mg, 0.014 mmol) in 02-saturated benzene at 25' for 16 hr led to complete consumption of 2. The products of the reaction, all of which were more polar than starting hydroperoxide, were reduced with NaBH4 (25 mg, 0.661 mmol in 10 ml of CN3OH) and then converted into the trimethylsilyl ethers for analysis by gas chromatography (GLC). Chromatography of the reduced and silylated product mixture on a 1%OV-17 column at 220' revealed a t least four volatile products with retention times expected for PGF-type products from a C-18 fatty acid.g Combined GLC-MS analysis of the product mixture indicates that two of these four compounds (retention times 9.5 and 10.5 min) are structurally analogous to authentic PGF1,. Prominent fragments of one of these compounds and PGF1, analyzed under identical GLC-MS conditions are presented in Table I.
Scheme I
COOH
\ OOH
HOO
o2
enzyme
fact, been concretely established by the isolation of two such compounds, PGG and PGH, from tissue homogenates which were actively engaged in the production of prostaglandim2 Recently, we reported a method for generating specific peroxy radicals from the corresponding hydroperoxide.3 This method is based on the fact that hydroperoxy hydrogens are readily abstracted by free radicals. Thus, hydroperoxides act as a source of the corresponding peroxy radical when treated with appropriate free radical initiators. We report here the application of this method to a hydroperoxide which should lead to a peroxy radical analogous to
Table I Mass Spectral D a t a (I.P. = 70 eV) of Trimethysilyl Ethers and Methyl Esters of PFG1, and Synthetic C-18 P G F (Retention Time 9.5 min) PGF 10
Polyunsaturated fatty acids are converted into hydroperoxides by a class of enzymes known as lipoxygenases (fatty acid: oxygen oxidoreductase). These enzymes have been identified in extracts from a variety of plant sources including soybeans: potato tubers: and wheat flour.6 With the soybean e n ~ y m e y-linolenic ,~ acid (all-cis-6,9,12-octadecatrienoic acid) could be converted into a 1:l mixture of the w-10 and a-6 hydroperoxides 2 and 3. The reaction was carried out on the ammonium salt of the fatty acid a t pH 7 for 2 min.8 The crude hydroperoxide fatty acids were then converted into the hydroperoxide methyl esters 2 and 3 by reaction with diazomethane. 2 and 3 could be separated by high performance liquid chromatography (HPLC) with 6 f t x in. of Corasil I1 and 0.24% isopropyl alcohol in hexane solvent. Thus, pure 2 could be isolated in an overall 15-20% yield based on starting fatty acid. The structures of 2 and 3
FragmenP
m/e
A
M
496 355 353 335 309 297 295
2 24 100 72
-
90
M - 233 M-251 M - 277 M-289
1.
C-18 PGF
80
62 50
Fragmenta
mte
94
M-90
468 3 55 325 307 281 269 295
2 48 72 61 75
MMMM-
233 251 277 289
100 22
Fragments between 200 and 500 m / e are shown and referred to the largest peak in this region as base. a
The mass spectra of the synthetic PGF's (which have carboxyl side chains smaller than PGFI, by CzH4, 28 m/e) are strikingly similar to PGFI,.~" Every fragment from the synthetic PGF's between 200 and 500 m/e with a relative intensity greater than 20% of base corresponds either t o a fragment of the same mass (loss of the carboxyl side chain) or to a fragment of m/e - 28 (retention of carboxyl side chain) from PGF1,.ll A mechanism consistent with our observations is presented in Scheme 11. tert- butoxy radicals generated by de-
3614
J. Org. Chem., Vol. 40, No. 24, 1975 3615
Communications composition of DBPO abstract the labile hydroperoxide hydrogen3 from 2 giving the peroxy radical. Cyclization of this radical to the endoperoxide followed by reductiod2 leads to PGF-type products. Scheme I1 COOCHB 2
+ t-BuO.
cyclization
C5H,,
-+
.o-0
/
bOH
+o,. +H.
OH
I
OH
4
The work reported here lends support to the notion that prostaglandin biosynthesis is a controlled free-radical reaction.' Other workers have noted the formation of prostaglandins in autoxidizing lipid.11J3 The method reported here has the advantage of producing specific peroxy radicals for study 8.8 compared to the rather random autoxidation format. Acknowledgmunt. This work was supported by grants from NIH and the Army Research Office (Durham). GLC-MS work was carried out a t the Research Triangle Institute. References a n d Notes (1) (a) 8. Samuelsson, Fed. Proc., fed. Am. SOC. Exper. B/ol., 31, 1442 (1972); (b) 8. Samuelsson, H. Granstrom, and M. Hamberg, In Nobel Symp. n2, 31-45 (1987): (c) D. H. Nugteren, R. K. Beerthuis, and D. A. van Dorp, lbld., 45-50 (1987). (2) (a) D. H. Nugteren and E. Hazelhof, Blochlm. Blophys. Acta, 328, 448 (1973); (b) M. Hamberg and B. Samuelsson, Proc. NaL Aced. Scl., USA, 70, 899 (1973); (c) M. Hamberg, J. Svenson, T. Wakabayashl, and B. Samuelsson, lbld., 71, 345 (1974). (3) (a) M. Funk, R. Isaac, and N. A. Porter, J. Am. Chem. Soc., 97, 1281 (1975); (b) see also J. A. Howard, Adv. Free-Radlcal Chem. 4, 49 (1972); (c) see also R. Hlatt, T. MIII, K. C. Irwln, and J. K. Castleman, J. Orp. Chem., 38, 1428 (1968). (4) (a) M. Hamberg and B. Samuelsson, Blochem. Blophys. Res. Commun., 21, 531 (1985); (b) M. Hamberg and 8 . Samuelsson, J. 8/01,Chem., 242, 5329 (1987); (c) M. Roza and A. Franke, Blochlm. Blophys. Acta, 318, 78 (1973); (d) J. P. Chrlstopher, E. K. Plstorlus, F. E. Regnler, and B. Axelrod, lbld., 289, 82 (1972), (5) T. Galllard and D. R.Phllllps, Blochem. J., 124, 431 (1971). (8) A. Graveland, Llplds, 8, 808 (1973). (7) Slgma Blochemlcal Co.; 45,500 unlts mg-l at pH 7.0. (8) N. A. Porter and M. 0,Funk, to be publlshed. (9) Several other more volatlle products were also observed. (10) See, for example, (a) C. J. Thompson, M. Los, and E. W. Horton, Llfe Sclences, 9, 983-988 (1970); (b) S. Nicosla and 01. Qalll, Anal. 810chem., 81, 192 (1974); (c) M. Hamberg, Eur. J. Blochem., 8, 135 (1968); (d) P. Wlodawer and B. Samuelsson, J. Blol. Chem., 248, 5873 (1973). (11) (a) Our mass spectra were compared to those of compounds obtalned by autoxidatlon of methyl a-llnolenate (W. A. Pryor and J. P. Stanley), Fragmentatlon patterns were generally slmllar wlth some exceptlons that may be ewpllcable by dlfferences In the proposed structures. We thank Professcr Pryor for this information before publlcation. (b) W. A. Pryor and J. P. Stanley, Abstracts of Papers, American Chemical Soclety Meeting, Chicago, Ill.,Aug 25-29, 1975, paper ORGN-50. (12) An alternate route to PGF compounds is via lsomerlzatlon of 4 to PGE's, followed by reductlon of PGE to PGF. (13) D. H. Nugteren, H. Vonkeman, and D. A. van Dorp, Recuell, 86, 1237 (1987).
Department of Chemistry Duke University .Durham. North, Carolina 27706 Received August 15, 1975
Ned A. P o r t e r * Max 0.F u n k
A Suggested Mechanism f o r t h e Production of Malonaldehyde d u r i n g the Autoxidation of Polyunsaturated F a t t y Acids. Nonenzymatic Production of Prostaglandin Endoperoxides d u r i n g Autoxidation
Summary: Autoxidation of methyl linolenate is shown to yield materials which give positive tests for both prostaglandin E and malonaldehyde, and it is suggested that both tests respond to prostaglandin-like endoperoxides which can be formed by autoxidation. Sir: When polyunsaturated fatty acids (PUFA) or esters containing three or more double bonds undergo autoxidation, a material is produced which develops color in a sensitive test with thiobarbituric acid (TBA).1-9 This TBA test is the most frequently used index of lipid peroxidation both in vitro and in vivo. Although the TBA-reactive material is frequently referred to as m a l 0 n a l d e h y d e , 5 ~it~has ~ ~ ~been ~~ known for some time3p4 that this material is predominantly nonvolatile; therefore, it is not malonaldehyde, but rather is a nonvolatile precursor of malonaldehyde. In 1962, Holman et aL4 suggested that a five-membered monocyclic peroxide is the nonvolatile malonaldehyde precursor. However, their mechanism does not appear to accommodate all the known facts of TBA-color production during PUFA autoxidation.1 A more attractive mechanism, in our view, irJ one in which the nonvolatile malonaldehyde precursor is a bicyclic endoperoxide analogous to that which is formed in the biosynthesis of prostaglandins.11-2a Figure 1 shows this mechanism as applied to methyl linolenate (18:3).Abstraction of an "internal" allylic hydrogen followed by reaction with 02 leads to peroxy radicals 4 and 5. Radical 4 has a structure which allows cyclization to endoperoxide radicals 9, which are allylic, probably via the oxy-bridged radicals 6. Radicals 9, then, can become converted into endoperoxides 10 or 11.21-23Radicals 4 and 5 also can lead to the conjugated hydroperoxides 7 and 8 which are known products of autoxidations. Our first indication that the nonvolatile malonaldehyde precursor is an endoperoxide came from comparisons of the responses of autoxidized solutions of 18:3to the TBA test and a test developed for prostaglandin E (PGE).1Q*z4~25 The PGE test, which involves rapid formation of absorption a t 278 nm upon addition of alcoholic b a ~ e , ~ ~probably * ~ ~ l ~isd relatively unspecific, but it is believed to convert PGE compounds into conjugated dienones such as PGB.24Since base is known to rapidly decompose secondary dialkyl peroxides to form ketones and alcohols,2e-2s we expected that endoperoxides, if produced in our autoxidations, would be converted by baae into PGE-type compounds. The PGEtype compounds would then react further with base to give PGB-type chromophores and a positive PGE test, It appeared reasonable a priori that endoperoxides could be formed nonenzymatically by autoxidation in our system since the suggested mechanism for their biosynthesis involves a radical cyc1ization;l la0 furthermore, Nugteren et al.2Dahave shown that autoxidation of 8,11,14-eicosatrienoic acid gives prostaglandins. Thus, we hypothesized that endoperoxides are produced on autoxidation of 18:3 and are the precursors of malonaldehyde under TBA-test conditions and PGB under conditions for the PGE test. (Note that 10 and 11 should give malonaldehyde but only 11 should give a PGE test.) Indeed, autoxidized 18:3 does give a PGE test. This is true regardless of whether the oxidation is spontaneous (Le., effected by pure air), or is initiated',' by ozone or NOr. On the other hand, 183, autoxidized under the same condi-
